RESEARCH ARTICLE Open Access

Cell type-specific expression profilingunravels the development and evolution ofstinging cells in sea anemoneKartik Sunagar1,2*†, Yaara Y Columbus-Shenkar1†, Arie Fridrich1, Nadya Gutkovich1, Reuven Aharoni1

and Yehu Moran1*

Abstract

Background: Cnidocytes are specialized cells that define the phylum Cnidaria. They possess an “explosive” organellecalled cnidocyst that is important for prey capture and anti-predator defense. An extraordinary morphologicaland functional complexity of the cnidocysts has inspired numerous studies to investigate their structure anddevelopment. However, the transcriptomes of the cells bearing these unique organelles are yet to be characterized,impeding our understanding of the genetic basis of their biogenesis.

Results: In this study, we generated a nematocyte reporter transgenic line of the sea anemone Nematostella vectensisusing the CRISPR/Cas9 system. By using a fluorescence-activated cell sorter (FACS), we have characterized celltype-specific transcriptomic profiles of various stages of cnidocyte maturation and showed that nematogenesis(the formation of functional cnidocysts) is underpinned by dramatic shifts in the spatiotemporal gene expression. Amongthe genes identified as upregulated in cnidocytes were Cnido-Jun and Cnido-Fos1—cnidarian-specific paralogs of thehighly conserved c-Jun and c-Fos proteins of the stress-induced AP-1 transcriptional complex. The knockdown of thecnidocyte-specific c-Jun homolog by microinjection of morpholino antisense oligomer results in disruption of normalnematogenesis.

Conclusions: Here, we show that the majority of upregulated genes and enriched biochemical pathways specific tocnidocytes are uncharacterized, emphasizing the need for further functional research on nematogenesis. The recruitmentof the metazoan stress-related transcription factor c-Fos/c-Jun complex into nematogenesis highlights the evolutionaryingenuity and novelty associated with the formation of these highly complex, enigmatic, and phyletically unique organelles.Thus, we provide novel insights into the biology, development, and evolution of cnidocytes.

Keywords: Cnidocyte transcriptome, Cnidocyst, Cnidaria, Nematostella vectensis, Cell type-specific transcriptome, Cnido-Jun

BackgroundCnidocytes, also known as stinging cells, are specializedneural cells that typify the phylum Cnidaria (sea anem-ones, corals, hydroids, and jellyfish) [1–3]. These cellscontain an organelle called cnida or cnidocyst, which isthe product of extensive Golgi secretions. Cnidae arecomposed of a complex capsule polymer characterized

by cysteine-rich peptides, such as minicollagens andnematocyst outer wall antigen (NOWA) [1, 4]. Further,the capsule elongates at its end into a tubule that ismade up of a polymer of peptides, including minicolla-gens, nematogalectins, and other structural proteins [5, 6].This tubule invaginates into the capsule during matur-ation and remains tightly coiled until activated during preycapture or defense, which results in the discharge of thecnidocyte capsule and uncoiling of the tubule. Cnidae arearguably the morphologically most complex organellesfound in nature, and their explosive discharge is one ofthe fastest biomechanical processes recorded in the animalkingdom [1, 7].

* Correspondence: ksunagar@iisc.ac.in; yehu.moran@mail.huji.ac.il†Kartik Sunagar and Yaara Y Columbus-Shenkar contributed equally to thiswork.1Department of Ecology, Evolution and Behavior, Alexander SilbermanInstitute of Life Sciences, The Hebrew University of Jerusalem, 9190401Jerusalem, IsraelFull list of author information is available at the end of the article

© Moran et al. 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Sunagar et al. BMC Biology (2018) 16:108 https://doi.org/10.1186/s12915-018-0578-4

Cnidocysts can be divided into three main categories: (i)nematocysts, the dart-shaped cnidae with spines on hol-low tubules that are used for prey piercing and venom in-jection; (ii) spirocysts, the elastic cnidae used for preyentanglement in Anthozoa; and (iii) ptychocysts which arefound exclusively in the cerianthid sea anemones. Nema-tocysts can be further divided into 25 subcategories basedon their morphology [2]. Such an extraordinary morpho-logical and functional complexity of cnidocysts has drivenseveral genetic and biochemical studies striving to identifythe molecular mechanisms underlying the formationof this enigmatic organelle [8–10]. However, no studyto date has uncovered the complete transcriptomiclandscape of cnidocytes, impeding our understandingof the evolutionary origin and the molecular mecha-nisms underpinning the maturation of these complexorganelles.Here, we characterize the cnidocyte transcriptome of

the sea anemone Nematostella vectensis, a rising modelorganism for the study of cnidarian biology [3, 11].Using the Clustered Regularly Interspaced Short Palin-dromic Repeats (CRISPR)/Cas9 system [12, 13], we gen-etically engineered transgenic lines that express afluorescent protein in their cnidocytes. We then proteo-lytically dissociated tissues and separated the resultingfluorescent cnidocytes from the other types of cells usinga fluorescence-activated cell sorter (FACS). This led tothe characterization and identification of thousands ofgenes that are differentially expressed at various stagesof cnidocyte maturation. Our findings shed light on theremarkable complexity and the dynamics of nematogen-esis (cnidocyst biogenesis). This experimental design fur-ther enabled the identification of cnidocyte-specifictranscription factors that have originated in Cnidariafrom a metazoan stress-related transcription factor com-plex around 500 million years ago. By genetically ma-nipulating Nematostella, we demonstrate the role ofthese unique transcription factors in nematogenesis, par-ticularly towards the formation of an integral capsuleprotein. Thus, our study provides novel insights into thefascinating biology and the development of cnidocytes—the first venom delivery apparatus to appear in animals.

ResultsGenerating a cnidocyte reporter lineTo generate a transgenic line that expresses a fluores-cent reporter protein in its cnidocytes, we used theCRISPR/Cas9 system to knock the reporter gene intothe genomic locus of the minicollagen NvNcol-3, whichis a major structural protein of the cnidocyst capsulewall and, hence, a well-characterized cnidocyte markerin Nematostella [14]. The reporter gene encodes mOr-ange2 [15] with a C-terminal RAS-derived membranetag (hereinafter referred to as memOrange2). We

followed a similar approach that was described before inNematostella [16], with the only exception being the ab-sence of promoter and regulatory sequences in thedonor construct (Fig. 1a). The microinjected embryosstarted expressing the fluorescent protein in their cnido-cytes 72 h post fertilization (hpf ) (Fig. 1b, c). Of 2633injected zygotes, 1053 survived the procedure, and ofthem, 314 exhibited by 72 hpf a signal detectable by afluorescent stereomicroscope. As expected of F0 Nema-tostella embryos, the transgene expression was restrictedto small-to-medium patches [17], a phenomenon thatwas also documented in primary polyps and juvenilestages (Fig. 1d, e). The second generation (F1) exhibitedspecific and strong expression of memOrange2 in a verywide population of cnidocytes at the planulae stage(Fig. 1f, g), with extremely strong expression in the ten-tacles of the primary polyp and juvenile stages (Fig. 1h,i). Our utilization of the CRISPR/Cas9 system exploitsthe homologous recombination-based DNA repairmechanism of the cell to insert the transgene into thenative gene and, thus, places the transgene under thecontrol of its native regulatory elements. Hence, the ex-pression pattern of the reporter gene accurately mir-rored the native expression of NvNcol-3, both at themRNA and protein levels, as demonstrated by doublefluorescent in situ hybridization (dFISH) and immuno-staining experiments (Fig. 1j–l), respectively. In the im-munostaining experiment, 85–100% of the cells werepositive for both antibodies (a mean overlap of 93%), in-dicating an overlapping expression. Only a small minor-ity of cells expressed either NvNcol-3 or memOrange2without an overlap.

Isolating distinct cnidocyte populationsWe dissociated tentacles of ten juvenile polyps of theNvNcol-3::memOrange2 transgenic line (3–4 months old;~ 50 tentacles in total) and sorted their cells accordingto the intensity of fluorescence. We initially collectedtwo different cell populations: one that included all cellsthat were positive for Hoechst stain (indication of an in-tact nucleus) and memOrange2 (hereinafter referred toas “positive cells”) and another population that was posi-tive for Hoechst but negative for memOrange2 (“nega-tive cells”) (Fig. 2a). Then, we employed a similarstrategy to collect the third population of cells that werenot only positive for Hoechst but expressed very highlevels of memOrange2 as well (“super-positive cells”)(Fig. 2c). Like before, we also collected negative cells forcomparison (Hoechst positive but memOrange2 nega-tive). Microscopic inspection of fluorescent cells revealedthat while the positive cells included many young devel-oping cnidocytes, which are characterized by small un-developed capsules or very few memOrange2-filledvesicles (Fig. 2b; Additional file 1: Figure S1), the

Sunagar et al. BMC Biology (2018) 16:108 Page 2 of 16

super-positive cell population was largely enriched withmature cnidocytes that were easily identifiable due to thepresence of elongated capsules (Fig. 2d; Additional file 1:Figure S1). To verify that memOrange2-positive cells areindeed young cnidocytes and that their fluorescence fol-lows the expression pattern of NvNcol-3, we fixed andco-immunostained dissociated cells. Indeed, these cellswere found to be positive for both memOrange2 andNvNcol-3 (Additional file 2: Figure S2). We isolatedmRNA from the aforementioned cell populations in tripli-cates and employed high-throughput sequencing tocharacterize gene expression patterns.

Characterizing maturation stage-specific cnidocytetranscriptomesTo characterize maturation stage-specific expression pro-files of cnidocytes, we employed two different compari-sons: (i) positives vs negatives and (ii) super-positives vsnegatives. The principal component analysis revealed dis-tinct gene expression profiles of the biological conditions(positives or super-positives vs their respective negatives)and uniform expression of biological replicates within the

respective condition (Fig. 3a, b). Differential expressionanalyses by two different methods in concert (DESeq2 andedgeR) enabled the stringent identification of a large num-ber of differentially expressed genes in positive andsuper-positive cell populations, in comparison to the nega-tive cells (Additional file 3: Figure S3). Our analyses revealedthe significant upregulation of genes encoding known cni-docyte markers, such as NvNcol-1, NvNcol-4, NEP-3,NEP-3-like, NEP-4, and NEP-5 [14, 18, 19] in positive cells(Additional file 4: Figure S4). Furthermore, certain genesthat are known to lack expression in cnidocytes [20–22],such as the neuronal markers FMRFamide and ELAV—thelatter only identified by DESeq2—were found to be down-regulated in both the positive and super-positive cells, incomparison to the negative cells (Additional file 4: FigureS4; Additional file 5: Figure S5). These results demonstratethe robustness of our experimental approach in isolatingcnidocytes and accurately characterizing their tran-scriptional profiles. However, failing to identifycnidocyte markers as differentially expressed in thesuper-positive cells, despite their upregulation inpositive cells, was surprising (Additional file 4: Figure

Fig. 1 Cnidocyte reporter line. a The depiction of the reporter construct used for the generation of the transgenic reporter line. b, c Expression ofmemOrange2 in planulae of F0 (injected) and F1 (first filial generation) (f–g). Magenta arrowheads point towards examples for positive animals,whereas yellow arrowheads point towards examples for negative animals. Scale bar is 250 μm. d, h Expression of memOrange2 in F0 and F1 primarypolyps. Scale bar is 250 μm. e, i Expression of memOrange2 in F0 and F1 juvenile polyps (2.5 months old). Scale bar is 1000 μm. j, k RNA expression andprotein localization of memOrange2 and NvNcol-3, revealed by dFISH and immunostaining, in early planulae and tentacles. l A higher magnificationcapture of immunostaining in a tentacle performed as in k. Scale bar in (j) is 100 μm, 50 μm in k, and 25 μm in l

Sunagar et al. BMC Biology (2018) 16:108 Page 3 of 16

S4; Additional file 5: Figure S5). Moreover, the tran-scripts encoding memOrange2 and NvNcol-3 were notidentified as significantly upregulated in both positiveand super-positive cnidocytes.A total of 1831 and 2149 differentially expressed genes

were identified in common by DESeq2 and edgeR inpositive and super-positive cell populations, respectively(Fig. 3, d; Additional file 6: Table S1). Overall, 613 and595 genes were commonly identified as upregulated ordownregulated, respectively, in the positive and super-positive cell populations, in comparison to the negativecells (Fig. 3e). Interestingly, 358 and 265 genes wereidentified as significantly upregulated and downregu-lated, respectively, in the positive cell population alone.These differentially expressed genes, identified only inthe positive cells, represent genes that are either upregu-lated or downregulated during the earlier stages of cni-docyte maturation but not in the mature cnidocyte-richsuper-positive population. Similarly, we identified 593and 348 genes as uniquely upregulated and downregu-lated, respectively, in the mature cnidocytes of thesuper-positive population but not in the earlier stages ofmaturation (Fig. 3e).

Spatiotemporal dynamics of cnidocyte transcripts andproteinsThe absence of significant enrichment of NvNcol-3 andmemOrange2 transcripts in both the positive and

super-positive cell populations was surprising. We sus-pected that this might be connected to the maturationstage of the isolated cnidocyte and that, perhaps, thetemporal differences in expression levels of memOrange2mRNA and protein may result in an inability to capturevery early-stage cnidocytes. To test this hypothesis, weperformed an ISH experiment, followed by immuno-staining. These combined assays enabled us to distin-guish between NvNcol-3 mRNA and protein expressionin the 3-day-old wild-type planulae (Fig. 4a–c) and be-tween the mRNA and the protein of the marker genememOrange2 in the transgenic planulae (Fig. 4d–f ).This experiment revealed spatiotemporal differences in

the mRNA and protein expression patterns of both thesegenes within cnidocytes. In both the wild-type andNvNcol-3 transgenic planulae, a significant proportion ofcnidocytes exclusively stained either at the mRNA levelor at the protein level (Fig. 4a–h). In seven wild-typeplanulae, we observed 117 cells that were positive onlyto NvNcol-3 RNA, 44 cells that were positive only forthe antibody, and 319 cells that were positive for both(Fig. 4g). In seven transgenic planulae, we detected 62cells that were positive only to memOrange2 RNA, 58cells that were positive only for the antibody, and 195cells that were positive for both (Fig. 4h). The cells thatonly expressed mRNA lacked capsules or contained un-developed capsules and multiple vesicles and are probablycnidoblasts or early-stage cnidocytes (Fig. 4), whereas

Fig. 2 Distinct cnidocyte populations. Panels a and c highlight the FACS-assisted sorting strategies employed in this study. The cells are first sorted intoHoechst-negative and Hoechst-positive populations, followed by the separation of the latter into memOrange2-positive, memOrange2-super-positive, andmemOrange2-negative populations. Panels b and d depict images of sorted positive and super-positive cells, as observed in differential interferencecontrast (DIC), mOrange, and DAPI filters, respectively, under a fluorescent microscope. A merger of these three images is also shown. Scale bar is20 μm in b and d

Sunagar et al. BMC Biology (2018) 16:108 Page 4 of 16

those containing only protein housed fully developed ornearly mature capsules (Fig. 4; Additional file 1: FigureS1). Only a subset of the cnidocytes, most of which werecharacterized by relatively small and immature capsules—typical of developing cnidocytes—exhibited expression atboth mRNA and protein levels (Fig. 4). These findings arein agreement with the recently reported co-localizationexperiments of NvNcol-3 [23].

Discovery of novel genes with cnidocyte-specificexpressionOne of the major objectives of this study was to identifynovel genes with cnidocyte-specific expression, as thiswould advance our knowledge regarding the cnidocyte

biology, their genetic basis of development, and evolu-tionary origin. First, to test the accuracy of our experi-mental and bioinformatic approaches for identifyingdifferentially expressed genes in cnidocytes, we choseten genes that were found in our current analyses to besignificantly upregulated in positive cells (log2 fold change inthe range of 2.8 to 5) (Fig. 5). Interestingly, nine of these tengenes were not previously described in the cnidocytes ofNematostella. They include genes that encode protein disul-fide isomerases (NVE15732, NVE15733, and NVE26200), ahomolog of the c-Jun transcription factor (NVE16876) thatwe named “Cnido-Jun,” a homolog of the c-Fos transcriptionfactor (NVE5133) that we named “Cnido-Fos1,” a highly de-rived homolog of NOWA (NVE17236), an M13 peptidase

Fig. 3 Differentially expressed genes within cnidocytes. Panels (a) and (b) show the clustering of biological replicates (memOrange2 negative:Neg1–3 and Neg5–7, positive: Pos1–3, and super-positive cell populations: Pos5–8) in a principal component analysis, where the axes representthe first two principal components, labeled with the percentages of variance associated with each axis. Panels (c) and (d) show MA-plots, whichrepresent the log ratio of differential expression as a function of the mean intensity for each feature. The total number of upregulated and downregulatedgenes, highlighted in the plot as blue and red circles, respectively, is also indicated. Panel (e) depicts a Venn diagram, showing a comparisonof differentially expressed genes identified in the positive and super-positive populations. The total number of genes uniquely identified aseither upregulated or downregulated in positive and super-positive cells can be attributed to early-stage and mature cnidocytes, respectively.Enriched and depleted annotation features of each of these cell populations are also indicated

Sunagar et al. BMC Biology (2018) 16:108 Page 5 of 16

(NVE13546), a calmodulin (NVE22513), an unknown proteincontaining thrombospondin and F5/8 type C domains(NVE5730), and a protein containing a galactose-binding lec-tin domain with an apparent homology to nematogalectins(NVE3843), a family of structural proteins found in cnido-cytes [6, 23]. The localization of expression using ISHhighlighted the cnidocyte-specific expression of these tengenes (Fig. 5). Unexpectedly, the expression pattern of indi-vidual genes exhibited a distinct spatiotemporal variability(Fig. 5). For example, the expression of the NOWA-likegene was limited to very few cnidocytes in the oral regionof the planula larva and the tentacles of the primary polyp,while the expression of Cnido-Jun was localized to both theoral and the central part of the planula, followed by an in-creased expression in the tentacles of the primary polyp(Fig. 5). The three protein disulfide isomerases also showednoticeably distinct expression patterns, strongly indicatingfunctional specialization of these enzymes (Fig. 5). Tovalidate the cnidocyte-specific expression of these novelgenes, we further conducted a double ISH experimentusing NvNcol-3 as a cnidocyte marker. An overlapping ex-pression was observed, demonstrating that these novelgenes, indeed, exhibit cnidocyte-specific expression profiles(Additional file 7: Figure S6).

Annotation of cnidocyte-associated biochemical pathwaysFunctional annotation of differentially expressed genes re-vealed that a large number of upregulated genes in cnido-cytes (46% and 36% in positive and super-positive

samples, respectively) are taxonomically restricted. Fur-ther, gene enrichment analyses enabled the identificationof functional categories that are significantly enriched ordepleted in positive and super-positive cell populations,providing novel insights into the biochemical pathways ofthese enigmatic cells (Fig. 3e; Additional file 8: Figure S7;Additional file 9: Figure S8). The enrichment of the termsrelated to the extracellular matrix (Fig. 3e; Additional file 8:Figure S7; Additional file 9: Figure S8) supports a strongevolutionary link between the structural components ofthe cnidocyst and the constituents of the extracellularmatrix [24]. Surprisingly, despite cnidocytes being special-ized neural cells known for expressing several ion channelsubtypes [25–27], a depletion of genes associated with ionchannels, such as “cation transport,” “sodium ion chan-nels,” and “potassium ion channels” was found (Fig. 3e;Additional file 8: Figure S7; Additional file 9: Figure S8).

Recruitment of a stress-response complex into nematogenesisBy retrieving homologs from the genomes and transcrip-tomes of a diversity of cnidarian species, we reconstructedthe phylogenetic histories of two transcription factors thatwere found to be specifically upregulated in cnidocytes(NVE16876 and NVE5133; Fig. 6). We named these tran-scription factors Cnido-Jun (4 and 1.5 log2 fold upregula-tion in positive and super-positive cells, respectively) andCnido-Fos1 (5 and 4.3 log2 fold upregulation in positiveand super-positive cells, respectively) to distinguish themfrom their paralogs encoding proto-oncogenic transcription

Fig. 4 Variability between the transcriptional and proteomic expression levels of cnidocyte-expressed genes. This figure shows that the expressionpatterns of NvNcol-3 and memOrange2 transcripts stained in FastRed (panels a and d) differ from the expression pattern of their protein products stainedwith Alexa fluor 488 (panels b and e). Examples for cells containing only RNA, only protein or both RNA and protein are indicated by magenta, gray andorange arrowheads, respectively (panels c and f). Quantifications of the cells positive for only RNA, only protein, or both RNA and protein (each from sevenanimals) are shown as column graphs (panels g and h). i A schematic diagram shows the overlap between the times of gene expression, proteinsecretion, and protein maturation. Scale bars are 10 μm

Sunagar et al. BMC Biology (2018) 16:108 Page 6 of 16

factors c-Jun and c-Fos. These transcription factors areknown to dimerize into an activation protein-1 (AP-1)complex, which is involved in various stress responses inBilateria and Cnidaria [28–30]. We show that Cnido-Junand Cnido-Fos1 protein-coding genes originated via geneduplication in the common ancestor of Hexacorallia—seaanemones and stony corals (Fig. 6a, b). However, only insea anemones, the c-Fos gene underwent an additionalround of duplication and led to the origination ofCnido-Fos2 (NVE23145) but was not identified as differen-tially expressed in a significant manner. Domain scanningof these transcription factors revealed cnidarian-specific in-sertions between the Jun and bZIP domains of the c-Junproteins—the latter is required for DNA binding (Fig. 6). Ina complete contrast, we found insertions in bilaterian c-Fosproteins that were missing in their cnidarian counterparts.

Interestingly, Cnido-Fos1 was the only protein that had dif-ferent residues than those implicated in the heterodimerformation (Fig. 6; Additional file 10: Figure S9), which maysuggest that this derived protein binds to novel partnerproteins.To understand whether Cnido-Jun, which exhibits

cnidocyte-specific expression, plays a significant rolein nematogenesis, we manipulated its expression inNematostella embryos. Using injection of morpholinoantisense oligo (MO), a well-established method inNematostella [31–33], we knocked down the expres-sion of Cnido-Jun and assayed cnidocyte developmentby immunostaining with α–NvNcol-3 at the planulastage (Fig. 7). Strikingly, animals injected with theCnido-Jun MO had far lower expression of NvNcol-3,and the cells positive for α–NvNcol-3 did not exhibit

Fig. 5 Novel genes with cnidocyte-specific expression. In situ hybridization expression patterns of novel cnidocyte-specific genes identified in thisstudy are shown in late planula and primary polyp. The elongated cells in the ectoderm, seen in the closeup image, can clearly be identified ascnidocytes according to the large unstained space, which is the cnidocyst capsule. Heatmaps of expression levels of each gene in positive (P) andsuper-positive (SP) cell populations, each across three technical replicates (positives and super-positive indicated as Pos1–3 and negatives asNeg1–3) are also provided. A color code for expression values, ranging from a gradient of red (downregulated) to blue (upregulated), is depictedat the bottom. Scale bars are 100 μm for planulae and primary polyps and 20 μm for closeup panels

Sunagar et al. BMC Biology (2018) 16:108 Page 7 of 16

the typical elongated shape observed in the controlMO animals (Fig. 7). Moreover, the very dense popu-lation of α–NvNcol-3-positive cells typically observedin the oral pole of the planulae was strikingly missingfrom the animals injected with Cnido-Jun MO (Fig. 7).To verify the specificity of the effect, we repeatedthese experiments with a second non-overlapping MO

against Cnido-Jun. We counted two different criteriathat represent normal cnidogenesis in Nematostella:(i) presence of high NvNCol-3 expression in aring-shaped domain around the oral pole of the pla-nula. In this count only one of 58 animals in theCnido-Jun MO-injected group exhibited expression inthis domain, whereas 43 of 74 animals exhibited this

Fig. 6 The deep origin of novel cnidocyte-specific transcription factors in Cnidaria. This figure depicts the phylogenetic histories of (a) the c-Junand b c-Fos family of proteins in Cnidaria. Duplication events leading to the origin of Cnido-Jun and Cnido-Fos1—the two cnidocyte-specifictranscription factors identified in this study (labeled in red font)—and c-Jun and c-Fos (indicated in green font) in Hexacorallia are shown in redcircles. Branches with node support ≥ 70 are shown in thick brown lines, and the bootstrap values for the major nodes are indicated. Domainorganization of these proteins are also depicted, where the Jun-like domain, the bZIP domains, and the c-Fos-related domains are indicated in blue,green, and purple, respectively. The DNA binding domain is marked by black arrows, while the dimerization domains are depicted in red bars

Sunagar et al. BMC Biology (2018) 16:108 Page 8 of 16

Fig. 7 Knockdown of Cnido-Jun results in decreased NvNcol-3 expression. (a) Morpholino antisense oligo design. The MO bases appear in purple,and the sequence of its binding site in the mRNA appears in green. The AUG translation start site is underlined. b Planulae that were injected aszygotes with control MO exhibit normal expression of NvNcol-3. c In contrast, planulae injected as zygotes with Cnido-Jun MO exhibit abnormalexpression of NvNcol-3, missing the dense expression domain in the aboral pole and most positive cells lack the typical elongated morphologyof NvNcol-3-expressing cells. Panels (d) and (e) present the effects of a second non-overlapping MO against Cnido-Jun and the control MO thatwas injected in parallel. Oral pole in panels (b–e) are indicated by asterisk, and scale bars are 50 μm. The numbers of embryos exhibiting anormal NvNcol-3 phenotype out of each injected group based on the two indexes described in the text appear below each merged picture

Sunagar et al. BMC Biology (2018) 16:108 Page 9 of 16

expression domain in the control MO-injected group,and (ii) presence of elongated NvNCol-3-positive cells thatmake at least 20% of all cells positive for this marker in theplanula. In this count, only one of 58 animals in the Cni-do-Jun MO-injected group exhibited such proportion,whereas 47 of 74 animals exhibited this proportion inthe control MO-injected group. For the secondnon-overlapping MO, the observed morphology andpercentage of affected animals counted by the same in-dices were highly similar, providing strong support forthe specificity of the assay (Fig. 7).

DiscussionAmong the large number of upregulated genes de-tected within the positive cell population (Fig. 3c, d;Additional file 3: Figure S3a), many were known cni-docyte marker genes, including toxin (NEP-3, NEP-3--like, NEP-4, and NEP-5) and structural protein-codinggenes (NvNcol-1 and NvNcol-4). At the same time,there was downregulation of certain neuronal markergenes (FMRFamide and ELAV) that are known to lackexpression in cnidocytes (Additional file 4: Figure S4;Additional file 6: Table S1). Though there was consistencywith regard to the downregulation of FMRFamide andELAV, we did not identify the upregulation of the afore-mentioned cnidocyte marker genes in the super-positivecell population (Additional file 5: Figure S5), which wasenriched with mature cnidocytes. This can be explainedby the fact that in mature cnidocytes, the capsule, which isa very tight polymer of various peptides and proteogly-cans, is already formed and the secretion of structural pro-teins is no longer required and is most probably wasteful.As a result, the expression of such genes diminishes withthe maturation of cnidocytes. This is clearly evident whenexamining the expression of memOrange2 across variouscell types. In our reporter line, memOrange2 was inte-grated into the genomic locus of NvNcol-3—the gene cod-ing a vital structural protein of the capsule (Fig. 1a). Theexpression of memOrange2 in the positive cell populationwas relatively higher than that in the negative cell popula-tion, but it dropped significantly in the super-positivepopulation, where it was identified as a highly downregu-lated gene (− 1.67 log2 fold change; p value 0.0005)(Additional file 4: Figure S4; Additional file 5: Figure S5).This clearly demonstrates that the expression of structuralproteins drops significantly with the maturation of cap-sules in the fully developed cnidocytes. Since the develop-ment of the mature capsule impedes the migration ofsecreted toxins, a similar pattern of expression can be ex-pected for toxin-coding genes at this stage of develop-ment. As explained in the following section, these resultscan also be attributed to the inherent dynamics of geneexpression across developing cnidocytes.

It should be noted that because the secreted memOr-ange2 protein requires 4.5 h at 37 °C for t1/2 maturation[34], a smaller fraction of cnidocytes in the very earlierstages of development may contain an immature non-glowing memOrange2 and, therefore, would get sorted intothe negative population. As a result, despite the increasedexpression in positive cells in comparison to the negativecells, memOrange2 was not identified as a differentiallyexpressed transcript. As explained in the “Results” section,the temporal variation in transcription and translation ofmemOrange2 could also contribute to this effect (Fig. 4i).However, the robustness of our experimental approach inisolating cnidocytes for generating cell type-specific tran-scriptomic profiles is strongly supported by multiple linesof evidence: (i) the significant upregulation of cnidocyte-specific markers (ranging between 1.8 and 4.6 log2 foldchange difference) and the downregulation of neuronalmarkers in positive cells (Additional file 4: Figure S4), (ii)microscopic examinations of isolated cells (Fig. 2b, d;Additional file 1: Figure S1), (iii) in situ hybridization exper-iments for the ten highly upregulated genes in cnidocytes(Fig. 5), and (iv) the functional annotation of upregulatedtranscripts (Additional file 8: Figure S7; Additional file 9:Figure S8). Moreover, the microscopic examination of thepositive cell populations revealed many cnidocytes in a veryearly stage of maturation (Additional file 1: Figure S1).These cells can clearly be seen to contain undevelopedcapsules and multiple secretory vesicles carrying prematurecnidocyst components, which indicate early stages ofnematogenesis when the capsule starts to form by massiveGolgi secretions [1]. Many were even completely missingcapsules but contained only memOrange2-filled vesicles(Additional file 1: Figure S1), suggesting that our approachcan capture cnidocytes in early stages of maturation. How-ever, the transgenic approach might have some limitationsin detecting genes expressed very early in cnidocyte devel-opment since the reporter fluorescence is by nature tem-porally lagging behind the expression of these genes (i.e.,the time required for memOrange2 translation and matur-ation). We speculate that this is the reason why somegenes previously reported to be involved in early nemato-genesis such as PaxA and Mef2 were not recovered in ouranalysis [23, 35].The dramatic difference we discovered in the expres-

sion profiles of certain genes in the positive andsuper-positive cell populations can be further attributedto the inherent dynamics of transcription and translationacross various stages of cnidocyte maturation. This wasrevealed by a combination of ISH and immunostainingexperiments, where we co-localized the transcripts andproteins encoded by memOrange2 or NvNcol-3 genesand is in agreement with previous results [23]. With thisapproach, we detected three different populations of cni-docytes: (i) cnidocytes in the very early stages of

Sunagar et al. BMC Biology (2018) 16:108 Page 10 of 16

maturation that only contained abundant transcripts butno proteins, suggesting that they were yet to develop acapsule; (ii) cnidocytes with high levels of both tran-scripts and proteins, suggesting that these were develop-ing cells but contained an immature capsule; and (iii)mature cnidocytes with fully developed capsules thatonly contained proteins and completely lacked tran-scripts for memOrange2 and NvNcol-3 genes (Fig. 4).This supports our hypothesis that transcription of toxinand structural protein-coding genes drastically falls inmature cnidocytes with the complete formation of thetightly polymerized capsule. Thus, we reveal spatiotem-poral dynamics of transcription and translation of cer-tain genes within cnidocytes (Fig. 4).Our experimental and bioinformatic approaches facili-

tated the identification of a large number of cnidocyte-specific genes, many of which were taxonomically restricted(46% and 36% in positive and super-positive samples, re-spectively; Additional file 6: Table S1), highlighting a pau-city of knowledge regarding the biology of cnidocytes. Weidentified nine genes that were significantly upregulated inpositive cells (log2 fold change in the range of 2.8 to 5) andwere not known to be specifically expressed in cnidocytes.The localization of expression in ISH experiments stronglysuggested that all ten genes exhibited cnidocyte-specific ex-pression (Fig. 5), albeit with a spatiotemporal variation. Forexample, three of these genes encoded protein disulfideisomerases (NVE15732, NVE15733, and NVE26200), whichprobably mediate the folding of cysteine-rich toxins andstructural proteins within cnidocytes that usually requirethe formation of multiple disulfide bridges for attaining theproper conformation and for carrying out their biologicalfunction [1, 24, 36]. Each of these enzymes exhibited dis-tinct expression profiles (Fig. 5), which is suggestive offunctional specialization of disulfide isomerases within thedifferent types of cnidocytes. Similarly, the expression ofthe NOWA-like gene was limited to a smaller populationof cnidocytes in the oral region of the planula and the ten-tacles of the primary polyp. In Hydra, NOWA has been im-plicated in the formation of nematocyst [37]. If this proteinserves a similar structural function in the cnidocytes ofNematostella, then the documented differences in expres-sion profiles are suggestive of distinct structural makeup ofdifferent cnidocyte populations.Interestingly, the expression of a calmodulin (NVE22513),

a highly conserved protein known to bind Ca2+ ions [38],was only restricted to certain regions of the tentacles in theprimary polyp (Fig. 5). Fluids of the mature cnidocytes areknown to contain high concentrations (ca. 500–600 mmol kg−1 wet weight) of Ca2+ ions that are bound bycertain proteins in the undischarged state [39], and the dis-sociation of Ca2+ ions from these partner proteins has beenimplicated in the rapid discharge of cnidocysts. Thus, therestricted expression profile of calmodulin points to the

functional specialization of Ca2+ ion-binding proteins in dif-ferent types of cnidocytes. Thus, the identification ofnumerous novel cnidocyte-specific genes has advanced ourknowledge regarding the biology and development ofcnidocytes.Our analyses enabled the discovery of Cnido-Jun and

Cnido-Fos1, two novel cnidocyte-specific transcriptionfactors that share very high sequence similarity with thec-Jun and c-Fos family of proteins. c-Jun and c-Fos areknown to constitute the AP-1 early-response transcrip-tion factors that are associated with stress, infection, cy-tokines, and other stimuli [29]. In Nematostella, a singlec-Jun (NVE21090) and a single c-Fos (NVE23145) havebeen identified as players in stress response [28, 40],whereas expression levels of Cnido-Jun and Cnido-Fos(referred to as Jun2 and Fos2, respectively, in one of theprevious studies) were found to be insensitive to inducedstress conditions. Results of our phylogenetic analysesrevealed that Cnido-Jun and Cnido-Fos1 genes originatedvia duplication events in the common ancestor of hexa-corallians (sea anemones and corals), approximately 500mya [41] (Fig. 6). These two genes were found to be sig-nificantly upregulated within cnidocytes, ranging be-tween 4 and 5 log2 fold upregulation, when compared tothe negative cells. However, unlike Cnido-Jun andCnido-Fos1, the expression of the c-Jun and c-Fos stressresponse transcription factors was not upregulated incnidocytes when compared to other cells. Therefore, it isvery likely that the cnidocyte-specific Cnido-Jun andCnido-Fos1 proteins, which might dimerize to form a“Cnido-Ap1” complex within cnidocytes, have been re-cruited into a function other than stress response, par-ticularly a function related to nematogenesis.To understand the role of these unique transcription fac-

tors within cnidocytes, we knocked down the expression ofCnido-Jun in Nematostella embryos. As this manipulationresulted in strong decrease in NvNcol-3 expression and de-fective morphology of the stained cells (Fig. 7), the resultsstrongly suggest that NvNcol-3, a gene encoding one of thebetter characterized nematocyst capsule components, is ei-ther a direct or indirect target of Cnido-Jun and its partnerprotein(s).The functional annotation of upregulated genes in posi-

tive and super-positive cell populations supported the earl-ier findings in Hydra showing that a significantly largeproportion of cnidocyte-specific genes in Cnidaria are taxo-nomically restricted [10]. Gene enrichment analyses facili-tated the discovery of significantly enriched or depletedannotation categories in the positive and super-positive cellpopulations, providing the first insights into the biochem-ical pathways of these peculiar cell types (Fig. 3e;Additional file 8: Figure S7; Additional file 9: Figure S8). Astrong enrichment of annotations related to the hydrolyticactivity was noted in the positive and super-positive cells,

Sunagar et al. BMC Biology (2018) 16:108 Page 11 of 16

which can be explained by the abundant presence ofproteolytic and hydrolytic enzymes in cnidocytes thatare responsible for processing cnidocyst structuralcomponents, as well as for exerting toxicity in prey orpredatory animals [8, 19, 42]. Annotation terms relatedto the extracellular matrix, such as the “basal lamina”(Fig. 3; Additional file 8e: Figure S7; Additional file 9:Figure S8), were found to be enriched in positive andsuper-positive cells. This supports the proposed evolu-tionary link between cnidocyst structural componentsand the extracellular matrix constituents [24]. Interest-ingly, even though cnidocytes are specialized neural cellsthat are known to express several subtypes of ion channels[25–27], a depletion of annotation terms related to ionchannels was identified (Fig. 3e; Additional file 8: FigureS7; Additional file 9: Figure S8). This can be explained bythe fact that many subtypes of ion channels are known tobe specifically expressed in certain other cell types, suchas neurons and muscles, but not in cnidocytes [27, 43–45]. The surprising enrichment of annotation terms like“cytokine-mediated signaling pathway,” “cellular responseto insulin stimulus,” and “regulation of cell motility” incnidocytes further highlights the presence of severaluncharacterized biochemical pathways that might be rele-vant to cnidocytes. Thus, several specialized ion channelsand biochemical pathways appear to be unique tocnidocytes.

ConclusionsWe conclude that nematogenesis or the differentiationof neuronal precursor cells into cnidocytes is a highlydynamic and multistep process that is accompaniedby complex shifts in the transcriptional and transla-tional profiles of differentiating cnidocytes. We high-light a drastic reduction in the transcriptional profilesof toxins and structural proteins within the maturecnidocytes that have a fully developed capsule. Weshow that a significantly large number of upregulatedgenes and biochemical pathways within cnidocytes areyet to be characterized, and thus, there is a remark-able paucity of knowledge regarding the biology ofcnidocytes. This study led to the identification ofnumerous novel protein-coding genes that showcnidocyte-specific expression. Most of them surpris-ingly exhibited spatiotemporal variation in expressionand indicated the presence of a large population ofuncharacterized cnidocytes. We also identified mul-tiple duplication events that led to the recruitment ofa bilaterian stress response complex into cnidocytesand how it is significant for nematogenesis. Thus, weprovide several fascinating insights into the biology,development, and the evolution of cnidocytes, whichare also the first venom-injecting cells to originatewithin animals, nearly 600 mya.

MethodsSea anemone cultureN. vectensis polyps were grown in 16‰ sea salt water at17 °C. Polyps were fed with Artemia salina nauplii threetimes a week. Spawning of gametes and fertilization wereperformed according to a published protocol [46]. Inbrief, temperature was raised to 25 °C for 9 h and theanimals were exposed to strong white light. Three hoursafter the induction, oocytes were mixed with sperm toallow fertilization.

TransgenesisFor generating a transgenic line expressing memOrange2under the native regulatory region of NvNcol-3, we micro-injected a Nematostella zygote with a mixture that includedguide RNA (250 ng/μl) of the sequence CAGUAGUUAGGGCAUCCCGG (as part of a transcript also carrying atracrRNA), Cas9 recombinant protein with nuclearlocalization signal (500 ng/μl; PNA Bio, USA), and a donorplasmid that includes two homology arms (spanning posi-tions 1,380,459-1,381,924 and 1,381,925-1,383,035 in scaf-fold 23 of the N. vectensis genome) spanning thememOrange2 gene. This construct encodes a chimeric pro-tein sequence that carries the first 32 amino acids ofNvNcol-3, fused to the memOrange2 sequence, separatedby a flexible linker sequence of 2 × (Gly-Gly-Ser) to allowproper folding of the fluorescent protein. The expression ofthe transgene in injected zygotes and embryos was moni-tored under a Nikon SMZ18 fluorescent stereomicroscopeequipped with a Nikon Ds-Qi2 monochrome camera andElements BR software (Nikon, Japan).

Dissociation of tentacles and cell sortingTentacles of Nematostella polyps were dissociated using acombination of papain (2 mg/ml; Sigma-Aldrich: P4762),collagenase (2 mg/ml; Sigma-Aldrich: C9407), and pro-nase (4 mg/ml; Sigma-Aldrich: P5147) in DTT (1.3 M)and PBS (10 mM sodium phosphate, 175 mM NaCl,pH 7.4). The tentacles were incubated with the proteolyticcocktail at 22 °C overnight along with Hoechst stain (Sig-ma-Aldrich) to stain the nucleus. This was followed by thedissociation of tissues into single cells by flicking the tubesgently and centrifugation at 400×g for 15 min at 4 °C. Thepellet was gently reconstituted in PBS. A small amount ofthe dissociated sample was subjected to microscopicexamination to verify successful cell separation, while theremaining sample was used for fluorescence-activated cellsorting in a FACSAria III (BD Biosciences, USA) equippedwith a 488-, 405- and 561-nm lasers and a 70-μm nozzle.Two distinct populations of cells were sorted andcollected: (a) Hoechst-positive and memOrange2-negative cell populations and (b) Hoechst-positiveand memOrange2-positive cell populations. The cellswere directly collected into TRIzol® LS reagent

Sunagar et al. BMC Biology (2018) 16:108 Page 12 of 16

(Thermo Fisher Scientific) in 3:1 reagent to sample ratio.The instrument was maintained at 4 °C throughout theprocedure. Positive and negative cells were collected fromthe tentacles of three independent batches of animals thatformed our biological replicates.

RNA isolation and sequencingTotal RNA from the positive and negative populationswas isolated using the TRIzol® LS reagent according tomanufacturer’s protocol. Following isolation, the RNAsamples were treated with Turbo DNase (ThermoFisher Scientific), followed by re-extraction with theTri-Reagent to remove DNase. RNA quality wasassessed on a Bioanalyzer Nanochip (Agilent, USA),and only samples with RNA integrity number (RIN) ≥8.0 were considered for sequencing (all samples exceptPOS7: 7.5 RIN). Sample library preparation for RNA se-quencing was accomplished using the Illumina TruSeqRNA library protocol (mean insert size of 150 bp). Thesamples were sequenced on Illumina Nextseq 500 highoutput v2 platform (2 × 40 bp), which generated anaverage of 50 million paired-end reads per replicate.The Illumina BaseSpace pipeline was used for de-multi-plexing and filtering high-quality sequencing reads.Additional quality filtering steps were performed usingTrimmomatic v0.36 [47] to remove adapters, leadingand trailing low-quality bases (below quality 3), veryshort reads (shorter than 20 bases), and reads with lessthan an average quality score of 20 using a sliding win-dow of 4 bases. The quality of the preprocessed and theprocessed data was verified using FastQC [48].

Differential gene expression and transcript annotationReads were aligned to the indexed genome of Nematostella[49] using STAR [50], followed by the quantification ofgene expression with HTSeq-count v0.6 [51]. TheNematostella gene models used in all analyses werepreviously utilized in other studies [52, 53] and can befound at https://figshare.com/articles/Nematostella_vectensis_transcriptome_and_gene_models_v2_0/807696. Differentialexpression analyses were performed using two Bioconductorpackages: DESeq v2.1 [54] and edgeR v3.16 [55]. Genesidentified in concert by these two methods were con-sidered as differentially expressed and were functionallyannotated using Blast2go v4.1 [56] against Toxprot [57]and Swissprot (May 16, 2014) databases. A cutoff of 50reads mapping to the negative sample, which filters outthe lowly expressed genes, was employed to increasethe stringency and accuracy of identifying differentiallyexpressed genes.

Whole mount in situ hybridization and immunostainingColorimetric ISH was performed following a publishedprotocol [58]. Double ISH combining nitro-blue

tetrazolium and 5-bromo-4-chloro-3′-indolyphosphate(NBT/BCIP) and FastRed (Roche, Germany) stainingwas also performed according to an established protocol[19]. Double fluorescent ISH (dFISH) was also per-formed according to published protocols [21, 22] withtyramide conjugated to Dylight 488 and Dylight 594fluorescent dyes (Thermo Fisher Scientific, USA). In ISHand FISH, embryos older than 4 days were treated with2u/μl T1 RNAse (Thermo Fisher Scientific) after probewashing, in order to reduce the background staining.Immunostaining was performed according to an estab-lished protocol [59], employing a commercially availablerabbit polyclonal antibody against mCherry (Abcam,USA) that cross-reacts with memOrange2 and acustom-made guinea pig antibody against NvNcol-3 [14],generously provided by Suat Özbek (Heidelberg Univer-sity). In experiments where ISH and immunostaining werecombined, the ISH staining was performed with FastRed(Sigma-Aldrich). Stained embryos, larvae, and adult polypswere mounted in either Vectashield antifade medium(Vector Laboratories, USA) or 85% glycerol and visualizedwith an Eclipse Ni-U microscope equipped with a DS-Ri2camera and an Elements BR software (Nikon, Japan) orwith a Leica SP5 upright confocal microscope equippedwith 405-, 488-, 561- and 594-nm lasers (Leica, Germany).

Immunostaining of dissociated Nematostella cellsImmunostaining protocol for single cells was adaptedfrom a published protocol for mammalian cell cul-tures [60]. PBS was added to the dissociated cellsfollowed by centrifugation at 400×g for 3 min at 4 °C.The pellet was gently reconstituted in PBS and placedon a coverslip covered with 2 μl of Cell-Tak (Corning,USA) inside a 12-well plate at room temperature.After 4 h in which the cell suspension was left for at-tachment, it was fixed by 4% paraformaldehyde for30 min at room temperature. The cell suspension waswashed in PBS and blocked by 5% BSA in PBS solu-tion for 20 min at room temperature. The primaryantibody staining was employed for 2 h at roomtemperature by a rabbit polyclonal antibody againstmCherry (Abcam) diluted 1:400. The primary anti-body was then washed by PBS, and a custom-madeguinea pig antibody against NvNcol-3 [14] diluted1:400 and DAPI (Thermo Fisher Scientific) diluted1:500 were added. The cell suspension was washedtwice with PBS and mounted by adding 8 μl Vecta-shield antifade medium (Vector Laboratories) on posi-tively charged slides. Details of all antibodies used inthis work are available in Table 1.

Morpholino antisense oligo (MO) assaysZygotes were microinjected with blocking-translationMOs of the sequences CTGCGATTCCTCCATTGGG

Sunagar et al. BMC Biology (2018) 16:108 Page 13 of 16

TTCTAA (Cnido-Jun MO I) or ACATAGTCTGAGTTTAATCCTCGCT (Cnido-Jun MO II) matching the 5′ un-translated region and the beginning of the coding se-quence of Cnido-Jun (Fig. 7a) and a control MO of thesequence CCTCTTACCTCAGTTACAATTTATA, whichis predicted to not bind to any Nematostella transcript.MOs were designed and supplied by GeneTools (USA).The MOs were injected at 0.9 mM, and the injection mixcontained 25 mM Dextran Alexa Fluor 488 (ThermoFisher Scientific). The MO assay followed a publishedprotocol [31]. The animals were fixed with ice-cold 4%paraformaldehyde for 45 min and incubated at 4 °C undergentle rotation, washed five times in PBS with 0.2% TritonX-100, and then stained with α–Ncol-3 following a pub-lished protocol [59].

Phylogenetic analysisHomologs of c-Jun and c-Fos transcription factors wereretrieved using the tblastn search [61] against NCBI’snon-redundant nucleotide sequence database, N. vecten-sis genome [49], the EdwardsiellaBase [62], and variouscnidarian transcriptomes [63]. The maximum likelihoodanalysis was utilized for the reconstruction of the mo-lecular evolutionary histories of the c-Jun and c-Fos pro-tein families. Trees were generated using PhyML 3.0[64], and node support was evaluated with 1000 boot-strapping replicates.

Additional files

Additional file 1: Figure S1. Maturation of cnidocytes. Images of thesorted positive and super-positive cells. The cells are observed indifferential interference contrast (DIC) and mOrange filter under afluorescent microscope. A merge of these two images is also provided.Scale bars are 10 μm. (TIF 6993 kb)

Additional file 2: Figure S2. Immunostaining of dissociated cells.Immunostaining with α-mCherry and α-NvNcol-3 of dissociated cells.Panels a, b, and c present single cells with vesicle-like structures positivelystained for mCherry but not in DIC. d, A spirocyte positive for both

antibodies. e, A nematocyst positive for both antibodies. Scale bars are10 μm. (TIF 6584 kb)

Additional file 3: Figure S3. Heatmap of differentially expressed genesin positive and super-positive cells. Heatmaps of the top 250 upregulatedand downregulated genes in positive (a) and super-positive (b) cnido-cytes, relative to negative cells, across technical replicates. A color codefor expression values (normalized log2 fold change rescaled between 2and − 2), ranging from a gradient of maroon (downregulated) to blue(upregulated), is also provided. (TIF 530 kb)

Additional file 4: Figure S4. Cnidocyte-specific markers in positive cells.a Differential expression values of cnidocyte and neuronal markers forpositive cells. b A heatmap of expression in the positive cell population,relative to negative cells, across technical replicates. A color code forexpression values, ranging from a gradient of maroon (downregulated)to blue (upregulated), is also provided. (TIF 2050 kb)

Additional file 5: Figure S5. Cnidocyte-specific markers in super-positivecells. a Differential expression values of cnidocyte and neuronal markers forsuper-positive cells. b. A heatmap of expression in the super-positive cellpopulation, relative to negative cells, across technical replicates. b A colorcode for expression values, ranging from a gradient of maroon(downregulated) to blue (upregulated), is also provided. (TIF 1903 kb)

Additional file 6: Table S1. Differentially expressed genes in positiveand super-positive cell populations. (XLSX 759 kb)

Additional file 7: Figure S6. Double in situ hybridization of novelgenes with NvNcol-3. Double in situ hybridization expression patterns ofthree novel cnidocyte-specific genes identified in this study are shown inlate planulae. The novel genes were stained by NBT/BCIP (brownish-pur-ple) while the NvNcol-3 marker transcript was stained by FastRed (red).Examples for overlapping cells are indicated by purple arrow heads; cellswhich express the assayed gene but do not overlap with the cnidocytemarker are indicated by green arrow heads. Scale bar is 100 μm. (TIF 12135 kb)

Additional file 8: Figure S7. Biochemical pathways in positivecnidocytes. a Enrichment of GO terms in positive cnidocytes. bUpregulated GO terms for biological processes, cellular components, andmolecular functions in the positive cell population. (TIF 813 kb)

Additional file 9: Figure S8. Biochemical pathways in super-positivecnidocytes. a Enrichment of GO terms in super-positive cnidocytes. bUpregulated GO terms for biological processes, cellular components, andmolecular functions in the super-positive cell population. (TIF 973 kb)

Additional file 10: Figure S9. Sequence alignment of c-Fos proteinfamily. Sequence identity is highlighted in shades of blue, while residuesimplicated in dimerization are marked by green arrowheads.Non-conserved dimerization residues in Cnido-Fos1 are shown in red.(TIF 951 kb)

AcknowledgementsWe are thankful to Dr. William Breuer (Interdepartmental Unit of the AlexanderSilberman Institute of Life Sciences) for his help with FACS experiments and to

Table 1 Antibodies used in this work

Name Manufacturer Catalog number Lot number RRID

Sheep anti-Digoxigenin-AP, Fab fragments Roche/Sigma 11093274910 13680324 AB_514497

Sheep anti-Fluorescein-AP, Fab fragments Roche/Sigma 11426338910 14973500 AB_514504

Sheep ant-Fluorescein-POD, Fab fragments Roche/Sigma 11426346910 10715122 AB_840257

Sheep anti-Digoxigenin-POD, Fab fragments Roche/Sigma 11207733910 11688900 AB_514500

Rabbit anti-mCherry antibody Abcam AB-ab167453 GR124924-35 AB_2571870

Guinea pig anti-NvNcol-3 Generously provided by Prof. Suat Özbek, HeidelbergUniversity

Custom polyclonal AB_2734725

Alexa Fluor 488 AffiniPure Donkey anti-Guinea Pig IgG (H+L) Jackson ImmunoResearch 706-545-148 118980 AB_2340472

Alexa Fluor 594-AffiniPure Goat anti-Rabbit IgG (H+L) Jackson ImmunoResearch 111-585-144 129771 AB_2307325

Sunagar et al. BMC Biology (2018) 16:108 Page 14 of 16

Dr. Naomi Melamed-Book (Imaging Unit of the Alexander Silberman Institute ofLife Sciences) for her help with confocal microscopy. We are grateful to the labof Yaron Shav-Tal (Bar-Ilan University) for their help with advice on single-cellimmunostaining. We are also thankful to Dr. Robert Zimmermann (University ofVienna) for his advice and support in bioinformatic analyses.

FundingThis work was supported by Israel Science Foundation grant no. 691/14, awardedto Y.M., and by the Marie Skłodowska-Curie Fellowship awarded to KS. KS was alsosupported by DBT-IISc Partnership Program.

Availability of data and materialsAll raw sequencing data generated in this project have been deposited to theSequence Read Archive (SRA) at the National Center for BiotechnologyInformation (Bioproject PRJNA391807; Biosamples SAMN07276326 andSAMN07276331 to SAMN07276341), https://www.ncbi.nlm.nih.gov/bioproject/. Allother data generated or analyzed during this study are included in this publishedarticle (and its supplementary information files). Gene models used in this studycan be found at https://figshare.com/articles/Nematostella_vectensis_transcriptome_and_gene_models_v2_0/807696.

Authors’ contributionsKS and YM designed the experiments. KS, YYCS, and YM performed theexperiments. KS performed bioinformatic analyses. KS, YYCS, and YM wrotethe paper. RA, AF, and NG assisted in some of the experiments. All authorsread and approved the final manuscript.

Ethics approval and consent to participateNot applicable.

Consent for publicationNot applicable.

Competing interestsThe authors declare that they have no competing interests.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims in publishedmaps and institutional affiliations.

Author details1Department of Ecology, Evolution and Behavior, Alexander SilbermanInstitute of Life Sciences, The Hebrew University of Jerusalem, 9190401Jerusalem, Israel. 2Evolutionary Venomics Lab, Centre for Ecological Sciences,Indian Institute of Science, Bangalore 560012, India.

Received: 29 May 2018 Accepted: 18 September 2018

References1. David CN, Ozbek S, Adamczyk P, Meier S, Pauly B, Chapman J, Hwang JS,

Gojobori T, Holstein TW. Evolution of complex structures: minicollagensshape the cnidarian nematocyst. Trends Genet. 2008;24(9):431–8.

2. Kass-Simon G, Scappaticci AA. The behavioral and developmentalphysiology of nematocysts. Can J Zool. 2002;80(10):1772–94.

3. Layden MJ, Rentzsch F, Rottinger E. The rise of the starlet sea anemoneNematostella vectensis as a model system to investigate development andregeneration. Wiley Interdiscip Rev Dev Biol. 2016;5(4):408–28.

4. Meier S, Jensen PR, Adamczyk P, Bachinger HP, Holstein TW, Engel J, OzbekS, Grzesiek S. Sequence-structure and structure-function analysis in cysteine-rich domains forming the ultrastable nematocyst wall. J Mol Biol. 2007;368(3):718–28.

5. Adamczyk P, Meier S, Gross T, Hobmayer B, Grzesiek S, Bachinger HP, HolsteinTW, Ozbek S. Minicollagen-15, a novel minicollagen isolated from Hydra, formstubule structures in nematocysts. J Mol Biol. 2008;376(4):1008–20.

6. Hwang JS, Takaku Y, Momose T, Adamczyk P, Ozbek S, Ikeo K, Khalturin K,Hemmrich G, Bosch TC, Holstein TW, et al. Nematogalectin, a nematocystprotein with GlyXY and galectin domains, demonstrates nematocyte-specific alternative splicing in Hydra. Proc Natl Acad Sci U S A. 2010;107(43):18539–44.

7. Nuchter T, Benoit M, Engel U, Ozbek S, Holstein TW. Nanosecond-scalekinetics of nematocyst discharge. Curr Biol. 2006;16(9):R316–8.

8. Balasubramanian PG, Beckmann A, Warnken U, Schnoelzer M, Schueler A,Bornberg-Bauer E, Holstein TW, Ozbek S. Proteome of the Hydranematocyst. J Biol Chem. 2012;287(13):9672-81.

9. Hwang JS, Ohyanagi H, Hayakawa S, Osato N, Nishimiya-Fujisawa C, Ikeo K,David CN, Fujisawa T, Gojobori T. The evolutionary emergence of cell type-specific genes inferred from the gene expression analysis of Hydra. ProcNatl Acad Sci U S A. 2007;104(37):14735–40.

10. Milde S, Hemmrich G, Anton-Erxleben F, Khalturin K, Wittlieb J, Bosch TC.Characterization of taxonomically restricted genes in a phylum-restrictedcell type. Genome Biol. 2009;10(1):R8.

11. Technau U, Steele RE. Evolutionary crossroads in developmental biology:Cnidaria. Development. 2011;138(8):1447–58.

12. Hsu PD, Lander ES, Zhang F. Development and applications of CRISPR-Cas9for genome engineering. Cell. 2014;157(6):1262–78.

13. Wright AV, Nunez JK, Doudna JA. Biology and applications of CRISPRsystems: harnessing nature’s toolbox for genome engineering. Cell. 2016;164(1–2):29–44.

14. Zenkert C, Takahashi T, Diesner MO, Ozbek S. Morphological and molecularanalysis of the Nematostella vectensis cnidom. PLoS One. 2011;6(7):e22725.

15. Shaner NC, Campbell RE, Steinbach PA, Giepmans BN, Palmer AE, Tsien RY.Improved monomeric red, orange and yellow fluorescent proteins derivedfrom Discosoma sp. red fluorescent protein. Nat Biotechnol. 2004;22(12):1567–72.

16. Ikmi A, McKinney SA, Delventhal KM, Gibson MC. TALEN and CRISPR/Cas9-mediated genome editing in the early-branching metazoan Nematostellavectensis. Nat Commun. 2014;5:5486.

17. Renfer E, Amon-Hassenzahl A, Steinmetz PR, Technau U. A muscle-specifictransgenic reporter line of the sea anemone, Nematostella vectensis. ProcNatl Acad Sci U S A. 2010;107(1):104–8.

18. Columbus-Shenkar YY, Sachkova MY, Macrander J, Fridrich A, Modepalli V,Reitzel AM, Sunagar K, Moran Y. Dynamics of venom composition across acomplex life cycle. Elife. 2018;7:e35014.

19. Moran Y, Praher D, Schlesinger A, Ayalon A, Tal Y, Technau U. Analysis ofsoluble protein contents from the nematocysts of a model sea anemonesheds light on venom evolution. Mar Biotechnol (NY). 2013;15(3):329–39.

20. Marlow HQ, Srivastava M, Matus DQ, Rokhsar D, Martindale MQ. Anatomyand development of the nervous system of Nematostella vectensis, ananthozoan cnidarian. Dev Neurobiol. 2009;69(4):235–54.

21. Nakanishi N, Renfer E, Technau U, Rentzsch F. Nervous systems of the seaanemone Nematostella vectensis are generated by ectoderm and endodermand shaped by distinct mechanisms. Development. 2012;139(2):347–57.

22. Wolenski FS, Layden MJ, Martindale MQ, Gilmore TD, Finnerty JR.Characterizing the spatiotemporal expression of RNAs and proteins in thestarlet sea anemone, Nematostella vectensis. Nat Protoc. 2013;8(5):900–15.

23. Babonis LS, Martindale MQ. PaxA, but not PaxC, is required for cnidocytedevelopment in the sea anemone Nematostella vectensis. Evodevo. 2017;8:14.

24. Ozbek S. The cnidarian nematocyst: a miniature extracellular matrix within asecretory vesicle. Protoplasma. 2011;248(4):635–40.

25. Bouchard C, Price RB, Moneypenny CG, Thompson LF, Zillhardt M, StalheimL, Anderson PA. Cloning and functional expression of voltage-gated ionchannel subunits from cnidocytes of the Portuguese Man O’War Physaliaphysalis. J Exp Biol. 2006;209(Pt 15):2979–89.

26. Li X, Martinson AS, Layden MJ, Diatta FH, Sberna AP, Simmons DK,Martindale MQ, Jegla TJ. Ether-a-go-go family voltage-gated K+ channelsevolved in an ancestral metazoan and functionally diversified in a cnidarian-bilaterian ancestor. J Exp Biol. 2015;218(Pt 4):526–36.

27. Moran Y, Zakon HH. The evolution of the four subunits of voltage-gatedcalcium channels: ancient roots, increasing complexity, and multiple losses.Genome Biol Evol. 2014;6(9):2210–7.

28. Elran R, Raam M, Kraus R, Brekhman V, Sher N, Plaschkes I, Chalifa-Caspi V,Lotan T. Early and late response of Nematostella vectensis transcriptome toheavy metals. Mol Ecol. 2014;23(19):4722–36.

29. Hess J, Angel P, Schorpp-Kistner M. AP-1 subunits: quarrel and harmonyamong siblings. J Cell Sci. 2004;117(Pt 25):5965–73.

30. Meng Q, Xia Y. c-Jun, at the crossroad of the signaling network. Protein Cell.2011;2(11):889–98.

31. Layden MJ, Rottinger E, Wolenski FS, Gilmore TD, Martindale MQ.Microinjection of mRNA or morpholinos for reverse genetic analysis in thestarlet sea anemone, Nematostella vectensis. Nat Protoc. 2013;8(5):924–34.

Sunagar et al. BMC Biology (2018) 16:108 Page 15 of 16

32. Magie CR, Daly M, Martindale MQ. Gastrulation in the cnidarianNematostella vectensis occurs via invagination not ingression. Dev Biol.2007;305(2):483–97.

33. Rentzsch F, Fritzenwanker JH, Scholz CB, Technau U. FGF signalling controlsformation of the apical sensory organ in the cnidarian Nematostellavectensis. Development. 2008;135(10):1761–9.

34. Shaner NC, Lin MZ, McKeown MR, Steinbach PA, Hazelwood KL, DavidsonMW, Tsien RY. Improving the photostability of bright monomeric orangeand red fluorescent proteins. Nat Methods. 2008;5(6):545–51.

35. Genikhovich G, Technau U. Complex functions of Mef2 splice variants in thedifferentiation of endoderm and of a neuronal cell type in a sea anemone.Development. 2011;138(22):4911–9.

36. Mouhat S, Jouirou B, Mosbah A, De Waard M, Sabatier JM. Diversity of foldsin animal toxins acting on ion channels. Biochem J. 2004;378(Pt 3):717–26.

37. Engel U, Ozbek S, Streitwolf-Engel R, Petri B, Lottspeich F, Holstein TW.Nowa, a novel protein with minicollagen Cys-rich domains, is involved innematocyst formation in Hydra. J Cell Sci. 2002;115(Pt 20):3923–34.

38. Friedberg F, Rhoads AR. Evolutionary aspects of calmodulin. IUBMB Life.2001;51(4):215–21.

39. Lubbock R, Gupta BL, Hall TA. Novel role of calcium in exocytosis:mechanism of nematocyst discharge as shown by x-ray microanalysis. ProcNatl Acad Sci U S A. 1981;78(6):3624–8.

40. Agron M, Brekhman V, Morgenstern D, Lotan T. Regulation of AP-1 byMAPK signaling in metal-stressed sea anemone. Cell Physiol Biochem. 2017;42(3):952–64.

41. Shinzato C, Shoguchi E, Kawashima T, Hamada M, Hisata K, Tanaka M, FujieM, Fujiwara M, Koyanagi R, Ikuta T, et al. Using the Acropora digitiferagenome to understand coral responses to environmental change. Nature.2011;476(7360):320–3.

42. Rachamim T, Morgenstern D, Aharonovich D, Brekhman V, Lotan T, Sher D.The dynamically evolving nematocyst content of an anthozoan, ascyphozoan, and a hydrozoan. Mol Biol Evol. 2015;32(3):740–53.

43. Grunder S, Assmann M. Peptide-gated ion channels and the simple nervoussystem of Hydra. J Exp Biol. 2015;218(Pt 4):551–61.

44. Gur Barzilai M, Reitzel AM, Kraus JE, Gordon D, Technau U, Gurevitz M,Moran Y. Convergent evolution of sodium ion selectivity in metazoanneuronal signaling. Cell Rep. 2012;2(2):242–8.

45. Jegla T, Marlow HQ, Chen B, Simmons DK, Jacobo SM, Martindale MQ.Expanded functional diversity of shaker K(+) channels in cnidarians is drivenby gene expansion. PLoS One. 2012;7(12):e51366.

46. Genikhovich G, Technau U. Induction of spawning in the starlet seaanemone Nematostella vectensis, in vitro fertilization of gametes, anddejellying of zygotes. Cold Spring Harb Protoc. 2009;2009(9):pdb.prot5281.

47. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illuminasequence data. Bioinformatics. 2014;30(15):2114–20.

48. Andrews S. FastQC: a quality control tool for high throughput sequencedata. 2010. Available online at: http://www.bioinformatics.babraham.ac.uk/projects/fastqc. Accessed 23 Sept 2018.

49. Putnam NH, Srivastava M, Hellsten U, Dirks B, Chapman J, Salamov A, TerryA, Shapiro H, Lindquist E, Kapitonov VV, et al. Sea anemone genome revealsancestral eumetazoan gene repertoire and genomic organization. Science.2007;317(5834):86–94.

50. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P,Chaisson M, Gingeras TR. STAR: ultrafast universal RNA-seq aligner.Bioinformatics. 2013;29(1):15–21.

51. Anders S, Pyl PT, Huber W. HTSeq—a Python framework to work with high-throughput sequencing data. Bioinformatics. 2015;31(2):166–9.

52. Moran Y, Fredman D, Praher D, Li XZ, Wee LM, Rentzsch F, Zamore PD,Technau U, Seitz H. Cnidarian microRNAs frequently regulate targets bycleavage. Genome Res. 2014;24(4):651–63.

53. Schwaiger M, Schonauer A, Rendeiro AF, Pribitzer C, Schauer A, Gilles AF,Schinko JB, Renfer E, Fredman D, Technau U. Evolutionary conservation of theeumetazoan gene regulatory landscape. Genome Res. 2014;24(4):639–50.

54. Love MI, Huber W, Anders S. Moderated estimation of fold change anddispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550.

55. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package fordifferential expression analysis of digital gene expression data.Bioinformatics. 2010;26(1):139–40.

56. Gotz S, Garcia-Gomez JM, Terol J, Williams TD, Nagaraj SH, Nueda MJ,Robles M, Talon M, Dopazo J, Conesa A. High-throughput functional

annotation and data mining with the Blast2GO suite. Nucleic Acids Res.2008;36(10):3420–35.

57. Jungo F, Bougueleret L, Xenarios I, Poux S. The UniProtKB/Swiss-Prot Tox-Prot program: a central hub of integrated venom protein data. Toxicon.2012;60(4):551–7.

58. Genikhovich G, Technau U. In situ hybridization of starlet sea anemone(Nematostella vectensis) embryos, larvae, and polyps. Cold Spring HarbProtoc. 2009;2009:pdb.prot5282.

59. Moran Y, Genikhovich G, Gordon D, Wienkoop S, Zenkert C, Ozbek S,Technau U, Gurevitz M. Neurotoxin localization to ectodermal gland cellsuncovers an alternative mechanism of venom delivery in sea anemones.Proc Biol Sci. 2012;279(1732):1351–8.

60. Kafri P, Hasenson SE, Kanter I, Sheinberger J, Kinor N, Yunger S, Shav-Tal Y.Quantifying beta-catenin subcellular dynamics and cyclin D1 mRNAtranscription during Wnt signaling in single living cells. Elife. 2016;5:e16748.

61. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignmentsearch tool. J Mol Biol. 1990;215(3):403–10.

62. Stefanik DJ, Lubinski TJ, Granger BR, Byrd AL, Reitzel AM, DeFilippo L, LorencA, Finnerty JR. Production of a reference transcriptome and transcriptomicdatabase (EdwardsiellaBase) for the lined sea anemone, Edwardsiella lineata,a parasitic cnidarian. BMC Genomics. 2014;15:71.

63. Kitchen SA, Crowder CM, Poole AZ, Weis VM, Meyer E. De novo assemblyand characterization of four anthozoan (phylum Cnidaria) transcriptomes.G3 (Bethesda). 2015;5(11):2441–52.

64. Guindon S, Dufayard JF, Lefort V, Anisimova M, Hordijk W, Gascuel O. Newalgorithms and methods to estimate maximum-likelihood phylogenies:assessing the performance of PhyML 3.0. Syst Biol. 2010;59(3):307–21.

Sunagar et al. BMC Biology (2018) 16:108 Page 16 of 16